import unittest from services.knowledge import extract_text, split_text class KnowledgeTextTest(unittest.TestCase): def test_extracts_utf8_text(self): self.assertEqual(extract_text("说明.txt", "你好,知识库".encode()), "你好,知识库") def test_split_text_keeps_overlap_and_all_content(self): text = "第一段。" * 250 chunks = split_text(text, chunk_size=120, overlap=20) self.assertGreater(len(chunks), 1) self.assertTrue(all(chunk for chunk in chunks)) self.assertLessEqual(max(map(len, chunks)), 120) def test_rejects_unsupported_binary_file(self): with self.assertRaisesRegex(ValueError, "暂仅支持"): extract_text("archive.zip", b"data") if __name__ == "__main__": unittest.main()